Purpose In many jurisdictions child witnesses who testify in court about their own sexual abuse are cross‐examined by a defence attorney. Children find this process to be distressing, and despite recent child‐focussed modifications to other aspects of the legal process, cross‐examination has remained largely unaltered. This lack of modification is due, in part, to the assumption that cross‐examination promotes truthful testimony (Wigmore, 1974 Evidence in trials at common law). However, little empirical research has investigated the effects of cross‐examination questions on children's reports of neutral and transgressive events. To examine these effects a laboratory‐based study was conducted. Method One hundred and twenty kindergarten (M = 6 years) and grade 2 (M = 8 years) students participated individually in a staged event. Children witnessed an adult commit a transgression and were then interviewed twice about it. Children first underwent a direct‐examination interview followed by either a direct‐ or cross‐examination interview. Results Children's reports of neutral events were significantly less accurate in Interview 2 cross‐examination, than they were in Interview 1 direct‐examination, whereas children interviewed twice with direct‐examination were equally accurate in Interviews 1 and 2. Furthermore, children whose second interview involved cross‐examination were less accurate in their reports of neutral events than were children whose second interview was a direct examination. Cross‐examination also affected some children's disclosures of a witnessed transgression. More of the older children provided truthful disclosures of the transgression in the initial direct examination compared with the Interview 2 cross‐examination. Conclusions Findings suggest that cross‐examination as used in this study may not be the most effective procedure for eliciting truthful testimony for both neutral and transgressive events from children aged between 5 and 8 years.